The overall aim is to understand the cognitive and neural bases of implicit probabilistic associative learning and how these vary with adult age. Learning is said to be implicit when people acquire information about structural regularities without intending to learn or becoming aware of what they have learned. Implicit learning is likely more important for adapting to the world than its explicit/declarative counterparts, particularly in later adulthood when little time is spent in formal schooling, a setting which emphasizes explicit learning. Nonetheless, the aging of implicit learning has been studied much less, perhaps because age-related deficits have sometimes been small or non-existent. This apparent sparing creates a paradox: Associative implicit learning calls on neural systems that show structural and functional declines in healthy aging, and so how would older people be relatively good at it? This is addressed by adopting themes from two research areas: Cognitive neuroscience provides evidence that two interactive learning systems--one based on the Medial Temporal Lobe (MTL) and the other based on the striatum--can contribute to such learning. The cognitive neuroscience of aging provides evidence that one of these neural systems is more sensitive to aging than the other, and that older brains often do the same task differently from younger brains. These themes lead to two hypotheses. First, that implicit probabilistic associative learning is not spared in healthy aging, but rather is characterized by persistent age-related learning deficits that can be detected as early as middle-age. These deficits are more pronounced as practice progresses, and cannot be attributed solely to age-related deficits in general processing resources. Rather, they reflect a fundamental deficit in the function of the striatal learning system. Second, the relative balance of these systems changes with age; early in training the MTL system dominates in both young and old adults, but as training proceeds the young adults come to rely on the striatal system, whereas the old continue to rely on their relatively intact MTL system. This enables old adults to maintain near- young levels of performance very early in training, but not later. These hypotheses will be tested by conducting behavioral and neuroimaging studies, using a recently developed triplets-learning-task (TLT). Aim 1 is to characterize age differences in implicit probabilistic associative learning. Aim 2 is to test behavioral implications for aging of the interactive systems hypothesis, and Aim 3 is to characterize the neural bases of implicit probabilistic associative learning in young and old adults, and relate them to behavior. These aims are important for fostering successful aging and independent living, because implicit associative learning is needed for adapting to new people, environments, and technologies, as well as for relearning skills after injury.